Requirements for PRPNet to replace LLRNet

[gd_barnes]

Originally Posted by rogue

Gary, could you restart the server with debuglevel=4 and e-mail me some files (debug.log, prpserver.log, prpserver.delay, and prpserver.ini) after you get that message on a client? I would also like to see the output from this select statement in the database:

select CandidateName, CompletedTests,
DecimalLength, LastUpdateTime,
k, b, c, n
from Candidate
where HasPendingTest = 0
and IsPRP = 0
and IsPrime = 0
and DoubleChecked = 0
and HasSierpinskiRieselPrime = 0
and DecimalLength > 0 ;

The combination of all of this data would be of great assistance to me.

Unfortunately we're having to wait on Max to load up the server because the 3.1.4 server is now dried and he needs to compile the programs for Linux for 3.1.5. Sorry. I can change the debug level but would need detailed instructions (as well as a correct input file format) for loading up a 3.1.4 version of the DB.

But...after much effort to figure out how to write this to a file and then give myself permissions to the MySQL folder, I've run the query. For the testing of k=2000-2050 on version 3.1.4, it selected 186993 rows. The full size file is 9.2 MB. Tarred up, it is 1.9 MB.

Since it is too big to post here, I'm posting a file that includes only k=2001-2003 tests. If you need the full file, let me know how you'd like it sent to you.

[gd_barnes]

Originally Posted by mdettweiler

lol. I was a tad busy yesterday but no, I actually didn't get the chance to start up 3.1.5 yesterday. It shouldn't be a problem though to do it today.

And no more "lol"'s please. It's not funny anymore.

[rogue]

Originally Posted by gd_barnes

Unfortunately we're having to wait on Max to load up the server because the 3.1.4 server is now dried and he needs to compile the programs for Linux for 3.1.5. Sorry. I can change the debug level but would need detailed instructions (as well as a correct input file format) for loading up a 3.1.4 version of the DB.

But...after much effort to figure out how to write this to a file and then give myself permissions to the MySQL folder, I've run the query. For the testing of k=2000-2050 on version 3.1.4, it selected 186993 rows. The full size file is 9.2 MB. Tarred up, it is 1.9 MB.

Since it is too big to post here, I'm posting a file that includes only k=2001-2003 tests. If you need the full file, let me know how you'd like it sent to you.

The file you attached will be fine.

The debuglevel setting is the last line of the prpserver.ini file.

You can load the database by either having a prpserver.candidates file and using the upgrade command line option or using prpadmin and an ABC file. Large ABC files tend will take a while to load through prpadmin.

[mdettweiler]

Originally Posted by gd_barnes

You said that yesterday. When today?

Here's the problem. 3.1.4 is showing lots of pairs but the server hasn't worked in nearly 2 days. I finally just stopped it. I just now restarted it and tried a client on it. It continues to say "No available candidates are left on this server." yet when I did the select command on the DB, it appeared that there are a ton of candidates above k=2009, which is the last k-value we were testing when the server just stopped.

Please don't just leave me/us hanging here. Thank you.

Ergh. I'm really sorry--I totally misplanned the timing of stuff. I've been rather busy in my offline life over the last couple of days and have barely had time to check the forum.

Okay, okay, I'll do it...right now!

BTW, if you check the server's web page, you'll notice that even though all those pairs are listed, they're listed as complete. I'm not sure if there's any way to clean them out eventually; Mark?

[mdettweiler]

Okay, I tried it again and...great, another compile error. Looks like it will have to wait a bit longer. I've emailed Mark about the error.

Gary, as I said in the note on humpford's desktop, I will be out of the house for much of today. I will nonetheless try to get to this when I get back, though I won't make any promises.

[rogue]

Originally Posted by mdettweiler

Ergh. I'm really sorry--I totally misplanned the timing of stuff. I've been rather busy in my offline life over the last couple of days and have barely had time to check the forum.

Okay, okay, I'll do it...right now!

BTW, if you check the server's web page, you'll notice that even though all those pairs are listed, they're listed as complete. I'm not sure if there's any way to clean them out eventually; Mark?

What do you mean by "complete"? Do you mean that they have been double-checked?

Eventually I will write a routine to remove candidates that need no further testing, but I don't know the best way to go about it. For example, do I wait a certain number of days after the test before removing it? Do I copy that data to another table instead of deleting it? Do I dump it to a file then delete it? I'm open to any ideas anyone has.

[Lennart]

Originally Posted by rogue

What do you mean by "complete"? Do you mean that they have been double-checked?

Eventually I will write a routine to remove candidates that need no further testing, but I don't know the best way to go about it. For example, do I wait a certain number of days after the test before removing it? Do I copy that data to another table instead of deleting it? Do I dump it to a file then delete it? I'm open to any ideas anyone has.

I think the best is to dump it to a file and delete them from DB.

Lennart.

[Mini-Geek]

Originally Posted by rogue

What do you mean by "complete"? Do you mean that they have been double-checked?

Eventually I will write a routine to remove candidates that need no further testing, but I don't know the best way to go about it. For example, do I wait a certain number of days after the test before removing it? Do I copy that data to another table instead of deleting it? Do I dump it to a file then delete it? I'm open to any ideas anyone has.

Perhaps add a few options to prpadmin, e.g. "Save completed tests to file" and "Clear completed tests from database". Whether this would require a double check to have been done or not (just a single test) could be decided by whether the server is set to double check candidates or not.

[mdettweiler]

Originally Posted by Mini-Geek

Perhaps add a few options to prpadmin, e.g. "Save completed tests to file" and "Clear completed tests from database". Whether this would require a double check to have been done or not (just a single test) could be decided by whether the server is set to double check candidates or not.

Right, that sounds like the way to go.

[rogue]

Originally Posted by Mini-Geek

Perhaps add a few options to prpadmin, e.g. "Save completed tests to file" and "Clear completed tests from database". Whether this would require a double check to have been done or not (just a single test) could be decided by whether the server is set to double check candidates or not.

That's possible. I do have some concerns though. If the server stats are ever reset, then the deleted tests will no longer be counted. They will still be counted for the user because those are not reset, but server_stats.html will be changed.

This is something along the lines of what I have thought of (previously) combined with what is being stated here, both done through prpadmin.

1) Allow delete by form. For example, specify 12*199^n+1 for some servers would delete all candidates and tests for numbers where k = 12, b = 199 and c = 1. The form would be dependent upon the servertype.

2) Allow delete by timestamp. For example, ask user the age criteria for deleting candidates and tests. prpadmin would ask for a number of days. The server would then delete all candidates with double-checks that are older than that number of days. One exception would be that the server would not delete the candidates and their tests if any tests/double-checks older than that are still pending. For example, if you want to delete everything older than 2 days, but there is a incomplete test from 3 days ago, then nothing would be deleted. This could be controlled by form. For a FixedKBC server, this would mean that an old outstanding test for k=111999 would not prevent the deleting of candidates for k=112001.

Thoughts?

[henryzz]

Originally Posted by rogue

Thoughts?

For 1) it would be helpful if it was possible to set an n range for the sequence you are removing.